Tiered storage for data processing

A method, comprising:

• performing, by one or more computing devices:
• receiving a processing request directed to a data set at a data processor that executes processing requests directed to the data set, wherein the data set and metadata describing the data set are maintained at a remote data store connected to the data processor over a network, wherein the data processor maintains data comprising one or more portions of the data set on local storage at the data processor for executing processing requests and the metadata describing the data set including a different portion of the data set not maintained at the local storage, and wherein the data processor is distinct from the remote data store;
• in response to receiving the processing request:
• identifying, by the data processor using the metadata, maintained at the data processor, describing the data set, the different portion of the data set not present at the data processor and used for executing the processing request, wherein the different portion of the data set not present at the data processor comprises other metadata describing one or more portions of user data in the data set;
• obtaining, by the data processor, the different portion of the data set, including the other metadata, from the remote data store;
• and
• responsive to the obtaining, executing, by the data processor, the processing request using the one or more portions of user data in the data set based, at least in part, on the obtained other metadata of the different portion of the data set.

A non-transitory, computer-readable storage medium, storing program instructions that when executed by one or more computing devices cause the one or more computing devices to implement:

• maintaining, at one or more storage devices at a data processor, one or more portions of a data set and metadata describing the data set, the data set fully maintained at a remote data store distinct from the one or more storage devices, the metadata describing a different portion of the data set not maintained at the one or more storage devices, and the data processor connected to the remote data store over a network;
• receiving, at the data processor, a request directed to the data set;
• in response to receiving the processing request:
• identifying, by the data processor using the metadata, maintained at the data processor, describing the data set, the different portion of the data set not present at the data processor and used for executing the processing request, wherein the different portion of the data set not present at the data processor comprises other metadata describing one or more portions of user data in the data set;
• obtaining, by the data processor, the different portion of the data set, including the other metadata, from the remote data store;
• and
• responsive to the obtaining, executing, by the data processor, the processing request using the one or more portions of user data in the data set based, at least in part, on the obtained other metadata of the different portion of the data set.